Evidence Log¶
Structured artifact — instantiates Affect–Evidence Separation
A durable written ledger that separates observations from interpretations, assumptions, and unknowns — and forces a column for what would prove the felt conclusion wrong.
Where a naming practice happens in the head, Evidence Log happens on the page. It is a persistent, inspectable artifact that sorts what you know from what you feel, assume, and don't yet know — literally into separate columns — and then adds the column reactivity always skips: what would count against the conclusion. Its defining trait is durability and accountability. Because it is written down, a second person can audit the reasoning, the log survives past the moment of feeling, and it can be reopened when a repeated or contested decision comes around again. It is not a decision and it does not name the affect for you; it is the ledger against which an affect-shaped claim is made to stand or fall in writing.
Example¶
At 2 a.m. an on-call site-reliability engineer is staring at a spike in checkout errors that started, she is certain, right after tonight's deploy. The certainty is doing real work: it wants her to roll back immediately and go to bed. Instead she opens the incident's Evidence Log. In the observations column go only timestamped facts: error rate up at 01:47; deploy landed 01:31; a database failover also fired at 01:44. In interpretation she writes her hunch — "the deploy caused it" — clearly flagged as a hunch, not a fact. In assumptions: "nothing else changed tonight." In unknowns: "was the failover routine or triggered?" Then she fills the counterevidence column — what she'd expect to see if the deploy were innocent — and notices the error spike began sixteen minutes after the deploy but within one minute of the failover. The log doesn't tell her what to do. But the felt certainty that had a rollback half-executed now has to survive a timeline it doesn't fit, and she pages the database team first.
How it works¶
- Separate by column, not by prose. Observations, interpretations, feelings, assumptions, and unknowns each get their own place. The physical separation is the mechanism; a paragraph blends them back together.
- Admit the feeling as data, in its own row. The felt certainty is logged, not banished — as a signal to explain, not a fact to assume.
- Force the counterevidence column. The log is not complete until it records what one would expect to observe if the affect-shaped claim were false, and whether that is present. This is the step that separates a log from a case file built to win.
- Leave unknowns visible. Empty cells are outputs too; a conspicuous "we don't know X" is what stops the feeling from silently filling the gap.
It is, in effect, a written defense against climbing the ladder of inference without noticing.[n1]
Tuning parameters¶
- Column schema — how finely observation is split from interpretation from assumption. More columns catch more substitution but raise the friction of keeping the log at all.
- Counterevidence rigor — from a single "what would change my mind?" line to a full disconfirmation checklist. The dial that decides whether the log tests the conclusion or merely documents it.
- Formality / auditability — private scratch ledger versus a shared artifact others review. Higher formality buys accountability at the cost of candor and speed.
- Retention and reopen cadence — one-off versus a living log revisited as new facts land. Essential for repeated or contested decisions; overkill for a one-time low-stakes call.
When it helps, and when it misleads¶
Its strength is accountability and memory: because the reasoning is external and separated, someone else can inspect it, the felt conclusion has to earn its place against written observations, and a later reviewer can see exactly where a feeling was doing an evidence's job. For high-stakes, contested, or repeated decisions this traceability is the whole point.
Its failure mode is that a log is only as honest as its counterevidence column. Run without one, it becomes a rationalized evidence search — a beautifully organized case file assembled to justify the conclusion the feeling already reached, confirmation bias with better formatting. It can also mislead through false completeness: a full-looking log implies the important evidence is in it, when the decisive fact may be an unknown nobody wrote down. And it adds real friction — logging every passing feeling is bureaucratic waste. The guarding discipline is to make the counterevidence and unknowns columns mandatory, not optional, and to reserve the full artifact for decisions whose stakes justify writing anything down.
How it implements the components¶
Evidence Log realizes the archetype's evidence-side machinery — the components that make a claim stand against something external:
evidence_check— the columns are the check: the affect-shaped claim is compared, in writing, against observable support, assumptions, and missing information.counterevidence_probe— the mandatory disconfirmation column actively records what would be true if the felt conclusion were wrong, and whether it is.
The log does not name the feeling for you — that affect_label and its judgment_claim come from Emotion Labeling — and it does not choose the action; the decision_boundary and reversible_action_option belong to Risk Evidence Review and Decision Pause Protocol. Its nearest twin is Nonclinical Coaching Evidence Check, which shares the evidence check but is a spoken, single-person script; the Evidence Log is a durable written ledger whose dedicated disconfirmation column the coaching script does not run.
Related¶
- Instantiates: Affect–Evidence Separation — the Evidence Log is the artifact that gives the archetype's "evidence has independent status" invariant a physical home.
- Consumes: Emotion Labeling supplies the named feeling and claim that the log then tests.
- Sibling mechanisms: Emotion Labeling · Decision Pause Protocol · Cognitive Reappraisal · Risk Evidence Review · Conflict De-escalation Review · Nonclinical Coaching Evidence Check
Editorial Notes¶
Form Classification¶
Form family: Record, Log & Register
Rationale: Evidence Log operates as a durable record, ledger, register, or trace whose value depends on preserving actual state or history because it a durable written ledger that separates observations from interpretations, assumptions, and unknowns — and forces a column for what would prove the felt conclusion wrong.
Independent corroboration: The frozen evidence defines Evidence Log as 'A durable written ledger that separates observations from interpretations, assumptions, and unknowns — and forces a column for what would prove the felt conclusion wrong', so its operative form is Record, Log & Register.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Organizational & Management Science
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Organizational learning practice developed the ladder-of-inference discipline of separating observable data from selected facts, meanings, assumptions, conclusions, and actions.
Related originating lineages:
- Ethnography & Qualitative Methods — Field-note and qualitative-research disciplines established the practice of separating observation from interpretation and reflexive judgment.
- Medicine & Healthcare — Clinical charting and differential-diagnosis records materially shaped durable separation of findings from interpretation.
- Psychology — Separating observations, interpretations, assumptions, and disconfirming evidence closely follows cognitive-behavioral self-monitoring and debiasing practice. Cognitive and clinical practice materially shaped explicit separation of observations, interpretations, assumptions, and falsifying evidence.
Review resolution: Argyris's organizational-learning lineage is the closest match to the log's observation-versus-interpretation columns. Psychological debiasing and qualitative field notes materially inform the generalized written ledger.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
Notes¶
[n1] The ladder of inference — the mostly invisible steps by which we move from raw observation to selected data to interpretation to conclusion to action, each rung feeling as solid as the one below. An evidence log works by making the lower rungs — what was actually observed versus what was inferred — sit in separate, visible columns. ↩